One or two preliminary remarks about this review-article may be in order. In New Left Review 58 (November-December 1969), Nicos Poulantzas wrote a very stimulating and generous review of my book The State in Capitalist Society; and in the following issue of nlr, I took up some of his comments and tried to meet some of his criticisms. This exchange attracted a good deal of attention, both in this country and elsewhere: obviously, and whether adequately or not, we had touched on questions concerning the state which Marxists and others felt to be important. I thought that the publication in English of Poulantzas’s own book on the state footnote¹ (it first appeared in French in 1968) would provide an opportunity to continue with the discussion that was then started, and to probe further some of the questions which were then raised. Unfortunately, the attempt to do this must, so far as I am concerned, be made in a much more critical vein than I had expected. The reason for this is that on re-reading the book in English five years after reading it in the original, I am very much more struck by its weaknesses than by its strengths. This is not a matter of poor translation: a random check suggests that the team of translators which was required for the task struggled valiantly and not unsuccessfully with an exceedingly difficult French text. It is a pity that the book is so obscurely written for any reader who has not become familiar through painful initiation with the particular linguistic code and mode of exposition of the Althusserian school to which Poulantzas relates. But too much ought not to be made of this: serious Marxist work on the state and on political theory in general is still sufficiently uncommon to make poor exposition a secondary defect—though the sooner it is remedied, the more likely it is that a Marxist tradition of political analysis will now be encouraged to take root.

Nor need a second and different objection that might be made against the book be taken as decisive, or even as particularly significant. This is its abstractness. The sub-title of the book in French (which the English edition does not reproduce) is: de l’Etat Capitaliste. But the fact is that the book hardly contains any reference at all to an actual capitalist state anywhere. Poulantzas says at the beginning of the work that ‘I shall also take into consideration not simply in research but also in exposition, concrete capitalist social formations’ (p. 24). But he doesn’t, not at least as I understand the meaning of the sentence. He seems to me to have an absurdly exaggerated fear of empiricist contamination (‘Out, out, damned fact’); but all the same, accusations of abstractness are rather facile and in many ways off the point—the question is what kind of abstractness and to what purpose. In any case, and notwithstanding the attention to concrete social formations promised in the above quotation, Poulantzas makes it quite clear that his main concern is to provide a ‘reading’ of texts from Marx and Engels, and also from Lenin, on the state and politics. Such a ‘reading’, in the Althusserian sense, is, of course, not a presentation or a collation of texts; nor is it a commentary on them or even an attempt at interpretation, though it is partly the latter. It is primarily a particular theorization of the texts. Poulantzas makes no bones about the nature of the exercise: ‘In order to use the texts of the Marxist classics as a source of information, particularly on the capitalist state’, he writes, ‘it has been necessary to complete them and to subject them to a particular critical treatment.’ footnote² Similarly, he notes that ‘these texts are not always explicit . . . Marx and Engels often analyse historical realities by explicitly referring to notions insufficient for their explanation. These texts contain valuable guide lines, so long as the necessary scientific concepts contained in them are deciphered, concepts which are either absent, or, as is more commonly the case, are present in the practical state’. footnote³ One may feel a bit uneasy about this ‘complementation’ of texts and at their subjection to ‘particular critical treatment’. But at least, the author appears to be playing fair in declaring what he is doing, and the enterprise is not in itself illegitimate—indeed, there is no other way of effecting a theorization. The question here too is how well the enterprise has been conducted, and whether the ‘deciphering’ has produced an accurate message. I will argue below that it has not and that much of Poulantzas’s ‘reading’ constitutes a serious misrepresentation of Marx and Engels and also of the actual reality he is seeking to portray.

I want to start by noting that the basic theme of the book, its central ‘problematic’, is absolutely right; and that Poulantzas, whatever else may be said about his work, directs attention to questions whose core importance not only for but in the Marxist analysis of politics cannot be sufficiently emphasized. What he is concerned to re-affirm is that the political realm is not, in classical Marxism, the mere reflection of the economic realm, and that in relation to the state, the notion of the latter’s ‘relative autonomy’ is central, not only in regard to ‘exceptional circumstances’, but in all circumstances. In fact, this notion may be taken as the starting-point of Marxist political theory. As with Althusser, ‘economism’ is for Poulantzas one of the three cardinal sins (the other two being ‘historicism’ and ‘humanism’); and even though his anti-‘economism’ is so obsessive as to produce its own ‘deviations’, there is no doubt that ‘economistic’ misinterpretations of the politics of classical Marxism have been so common among enemies and adherents alike that even some stridency in the assertion of the central importance of the concept of the relative autonomy of the political in Marxist theory may not come amiss. footnote⁴

Still, to insist on this is only a starting-point, however important. Once it has been established, the questions follow thick and fast: how relative is relative? In what circumstances is it more so, or less? What form does the autonomy assume? And so on. These are the key questions of a Marxist political sociology, and indeed of political sociology tout court. It would be absurd to blame Poulantzas for not having, in this book, provided an answer to all these questions. The real trouble, as I see it, is that his approach to these questions prevents him from providing a satisfactory answer to them. In my Reply to Poulantzas in nlr 59, I said that his mode of analysis struck me as leading towards what I then called ‘structural super-determinism’. I think that was right but that a more accurate description of his approach and of its results would be structuralist abstractionism. By this I mean that the world of ‘structures’ and ‘levels’ which he inhabits has so few points of contact with historical or contemporary reality that it cuts him off from any possibility of achieving what he describes as ‘the political analysis of a concrete conjuncture’. footnote⁵ ‘Everything happens’, he writes, ‘as if social classes were the result of an ensemble of structures and of their relations, firstly at the economic level, secondly at the political level and thirdly at the ideological level.’ footnote⁶ But even if we assume that classes are the product of such an ‘ensemble’, we want to know the nature of the dynamic which produces this ‘ensemble’, and which welds the different ‘levels’ into the ‘ensemble’. Poulantzas has no way that I can discern of doing this: the ‘class struggle’ makes a dutiful appearance, but in an exceedingly formalized ballet of evanescent shadows. What is lacking here is both any sense of history or for that matter of social analysis. One example is Poulantzas’s treatment of the notion of ‘class-in-itself’ and ‘class-for-itself’. These are described as ‘1847 formulae’ of Marx, which ‘are merely Hegelian reminiscences. Not only do they fail to explain anything, but they have for years misled Marxist theorists of social classes’. footnote⁷ But what, it may then be asked, is to take the place of these ‘Hegelian reminiscences’, since we clearly do need some means of tracing the dynamic whereby a class (or a social aggregate) becomes an ‘ensemble’ in which the economic, the political and the ideological ‘levels’ achieve the necessary degree of congruence?

Poulantzas sees the problem: ‘A class’, he says, ‘can be considered as a distinct and autonomous class, as a social force, inside a social formation, only when its connection with the relations of production, its economic existence, is reflected on the other levels by a specific presence’. footnote⁸ Leaving aside this oddly ‘economistic’ reflectionism, after so much denunciation of it, one must ask what is a ‘specific presence’? The answer is that ‘this presence exists when the relation to the relations of production, the place in the process of production, is reflected on the other levels by pertinent effects.’ footnote⁹ What then are ‘pertinent effects’? The answer is that ‘we shall designate by “pertinent effects” the fact that the reflection of the place in the process of production on the other levels constitutes a new element which cannot be inserted in the typical framework which these levels would present without these elements’. footnote¹⁰ This might be interpreted to mean that a class assumes major significance when it makes a major impact upon affairs—which can hardly be said to get us very far. But Poulantzas does not even mean that. For he also tells us, ‘the dominance of the economic struggle’ (i.e. ‘economism’ as a form of working-class struggle—R.M.) does not mean ‘an absence of “pertinent effects” at the level of political struggle’—it only means ‘a certain form of political struggle, which Lenin criticizes by considering it as ineffectual’. footnote¹¹ So, at one moment a class can only be considered as distinctive and autonomous if it exercises ‘pertinent effects’, i.e. a decisive impact; next moment, these ‘pertinent effects’ may be ‘ineffectual’. Poulantzas never ceases to insist on the need for ‘rigorous’ and ‘scientific’ analysis. But what kind of ‘rigorous’ and ‘scientific’ analysis is this? Indeed, what kind of analysis at all?

I now want to return to the issue of the relative autonomy of the state and show how far Poulantzas’s structuralist abstractionism affects his treatment of the question. Not only does his approach seem to me to stultify his attempt to explain the nature of the state’s relationship to the dominant class: it also tends to subvert the very concept of relative autonomy itself. Driven out through the front door, ‘economism’ reappears in a new guise through the back. Thus, Poulantzas tells us that ‘power is not located in the levels of structures, but is an effect of the ensemble of these levels, while at the same time characterizing each of the levels of the class struggle’. footnote¹² From this proposition, (which strikes me as extremely dubious, but let it pass), Poulantzas moves on to the idea that ‘the concept of power cannot thus be applied to one level of the structure. When we speak for example of state power, we cannot mean by it the mode of the state’s articulation at the other levels of the structure; we can only mean the power of a determinate class to whose interests (rather than to those of other social classes) the state corresponds’. footnote¹³ Now this, I should have thought, is manifestly incorrect: it is simply not true that by ‘state power’, we can only mean ‘the power of a determinate class’. For this, inter alia, is to deprive the state of any kind of autonomy at all and to turn it precisely into the merest instrument of a determinate class—indeed all but to conceptualize it out of existence. Lest it be thought that I exaggerate, consider this: ‘The various social institutions, in particular the institutions of the state, do not, strictly speaking, have any power. Institutions, considered from the point of view of power, can be related only to social classes which hold power’. footnote¹⁴

As if uneasily aware of the implications of what he is saying, Poulantzas assures us that ‘this does not mean that power centres, the various institutions of an economic, political, military, cultural, etc, character are mere instruments, organs or appendices of the power of social classes. They possess their autonomy and structural specificity which is not as such immediately reducible to an analysis in terms of power.’ footnote¹⁵ This half-hearted concession does not dissipate the confusion: it only compounds it. The reason for that confusion, or at least one reason for it, is Poulantzas’s failure to make the necessary distinction between class power and state power. State power is the main and ultimate—but not the only—means whereby class power is assured and maintained. But one of the main reasons for stressing the importance of the notion of the relative autonomy of the state is that there is a basic distinction to be made between class power and state power, and that the analysis of the meaning and implications of that notion of relative autonomy must indeed focus on the forces which cause it to be greater or less, the circumstances in which it is exercised, and so on. The blurring of the distinction between class power and state power by Poulantzas footnote¹⁶ makes any such analysis impossible: for all the denunciations of ‘economism’, politics does here assume an ‘epiphenomenal’ form.